Optimum placement of capacitor in distribution system using a DVR with ANN Technique S Tanya Priyanka*, J Krishna Kishore** *1 M.Tech student, Department of E.E.E, QIS College of Engineering and Technology, Ongole, India **2 Asst. Professor, Department of E.E.E, QIS College Of Engineering and Technology, Ongole, India Abstract This paper reports concerns of capacitor placement in power distribution systems in the locality of a dynamic voltage restorer (DVR). A feedback controlled simulation model of an actual DVR is developed to perform case studies. Cogency of this model is established by associating the waveforms from the model with field measurements. Issues relating to expansion of a DVR in the locality of power factor correction capacitors are analyzed. Some considerations regarding the placement of capacitors in the vicinity of a DVR are drawn based on the simulation models and a novel Artificial Neural Network Based converter had been introduced in the DVR system and location of capacitor is justified. Index Terms ANN, Capacitor, DVR, VSI. distribution power quality. The DVR pays series voltage boost technology using solid state switches for compensating voltage sags/swells. Usually voltage source inverter technology is used. The DVR solicitations are mainly for sensitive loads that may be drastically affected by fluctuations in system voltage. The basic concept of a DVR is shown in Fig. 1. DVR technology has been commercialized for portion sensitive loads. The operation, power electronic requirements, and topologies of DVRs are described in Refs. [5 8, 10]. The series voltage boost technology is described using phasors in Ref. [2]. I. INTRODUCTION The application of voltage-sensitive equipment, such as automatic production lines, computer centers, hospital equipment, programmable logic controllers (PLC), adjustable speed drives (ASD), and airconditioning controllers [1, 2], has been increasing. Voltage sag is defined as the reduction in voltage RMS between 0.1 and 0.9 PU within 0.5 cycles to a few seconds [3]. Swell is defined as an increase in nominal voltage between 1.1 and 1.8 PU during 0.5 cycles to 1 minute. Faults or large induction motors starting in the power system may cause voltage sags or swell. Consequently, other equipment may shut down [3, 4]. A solution for power quality improvement is to use custom power devices like a dynamic voltage restorer (DVR). External energy storage is necessary to provide the requirement for real power. Thus, the maximum amount of real power that can be provided to the load during voltage sag mitigation is a deciding factor of the capability of a DVR. However, the energy requirement cannot be met by the application of such phase advance technology alone to compensate the deep sag of long duration; in addition, since there are the limitations in the provider of these energy devices, it is necessary to minimize energy injection [5]. DVRs area class of custom power devices for given that reliable Fig. 1. Components of a generic DVR An actual DVR installation in was facing control interference issues due to the switching of power factor correction banks in the vicinity of the DVR. As a result of transient overvoltages produced due to the switching event, the power electronic device by passed into protective operation mode thus, rendering the load fro m maximum utilization of the capability of the DVR. Hence, alternative placements of capacitors in the vicinity of the DVR were probed for a solution. The organization of this paper includes sections on the dynamic voltage restorer, the description of an actual operating DVR and its controls, a simulation model and validation, followed by a detailed discussion on the interaction between a DVR and capacitors and case 219 www.ijaegt.com
studies to draw some considerations on capacitor placement strategies. far from the final optimum. The LM algorithm is the best by comparing GN algorithm and GD method. II. THE DVR AND ANN CONTROLLER The proposed DVR system composed of VSI, LC filter, Energy storage, controller and an injection Transformer as shown in figure 1. The VSI inject an appropriate voltage to restore a sensitive load voltage through an injection transformer and LC filter from an external energy storage unit. Control strategy is the main part of the DVR system. The main function of a DVR control system is to detect the disturbances occurring in the system and compute the missing voltage to generate gate pulses using Discrete PWM generator, then the IGBT inverter converts the input DC voltage to a sinusoidal AC voltage through an LC filter and injection Transformer. The compensating voltage injected by the DVR system is stopped, only after the absence of the sag disturbance. In this work, the park s transformation is used to calculate the missing voltage. The abc to dqo transformation is transformed the 3Ф stationary coordinate system to dqo rotating coordinate system. In abc_to_dq0, the following transformation is used. In this paper, an ANN controlled DVR is modelled for PQ enhancement in a three-phase, three wire, power distribution system. A three leg VSI is used to inject or absorb the appropriate voltage through an LC filter and an injection transformer to compensate load voltage from the distorted supply voltage. Initially the data from PI controller is stored in workspace. These saved data is trained offline using ANN. In addition to that, the DVR is also used to protect the sensitive linear load. The simulation results show the effectiveness of the voltage restoration and its performance investigation of both control techniques. Fig: 1.1 ANN Controller III. SIMULATION MODEL Table: 1 System Parameters Supply voltage peak (phase to phase) 12.47 kv Desired load voltage peak (phase to 0.97 * 12.47 kv phase) Load power factor 0.86 DC voltage link 1.8 MJ cap bank Switching frequency of PWM 3 khz Cut off frequency of low pass filter 1 khz Injection transformer 6 MVA PWM side winding [V peak (kv), R (pu), L (pu)] 2.828, 0.0064, 0.0429 SER side winding[v peak (kv), R (pu), L (pu)] 2.513, 0.00815, 0.0544 Three phase load served 11.62 MVA (10 MW) Shunt capacitor banks for power factor 1.8 MVAr correction The simulation methodology used in this case study comprises the following aspects: modeling using available data and approximation, validating the computer model by comparing results of simulation against field measurements from actual A. ANN Controller To improve the performance of the compensating device, a multilayer back propagation type ANN controller is used. The matlab toolbox is used to train ANN. The training algorithm used in the ANN controller is Levenberg Marquardt Back propagation algorithm. Gradient Descent (GD) Method is the first order optimization algorithm and it is used to find a local minimum of a given function. This method is robust when it starts far of the final minimum; however it has poor final convergence. The LM back propagation algorithm is the second order optimization and it interpolates between the GD and the Gauss Newton (GN) algorithm. The LM algorithm is more robust and it finds a solution even if it does begin very Fig. 2. Voltage sag event recorded on the supply side of the actual DVR installation 220 www.ijaegt.com
physical installation, and finally utilizing the validated computer model to perform what if scenarios. The computer models werecreated using the SimPowerSystems blockset of MatLab Simulink [3]. The simulation model of the system under consideration consists of the following components: a programmable voltage source that models the sub-transmission side of the system with the capacity to inject harmonics and sag/swells; source impedance modeled with the system short circuit ratio (SCR); sub-transmission to distribution voltage level step down transformer; three phase series injection transformer in series with the supply and the load; three phase load with power factor correction banks; DVR subsystem consisting of energy link, PWM based inverter, and LC filters; and dq0 transform based feedback controller. The dq0 transform based feedback controller is similar to the one described in [7]. The simulation model is validated by running several test cases similar to those experienced by the actual DVR installation and comparing the results from the model to the measurements obtained from the field. A test case is described as follows: a voltage sag of depth 30% lasting five cycles, recorded at one of the primary buses of the actual DVR installation, shown in Fig. 2, is imposed on the simulation model from 0.3 to 0.3833 s. Fig. 3a and b show the response of the actual DVR installation and simulation model, respectively. It is evident that the voltage waveform obtained from the computer simulation possesses higher order harmonics than observed in the field. This is explained by the use of a generic low pass filter with cut off frequency near 1 khz (approximately 16th order on a 60 cycle wave). The choice of the cut off frequency for the low pass filter was dictated by general rules of thumb in the absence of actual data pertaining to the low pass filter at the DVR installation. IV. INTERACTION BETWEEN A DVR AND POWER FACTOR CORRECTION CAPACITOR BANKS actions with inductances and other capacitors in the system. This is a simple consequence of the analysis of an RLC circuit which generally has complex poles: the imaginary parts of the complex poles are the resonant frequencies [1]. The impedance function and the magnitude of the impedance of the series RLC circuit are. In the absence of feedback or active components, these oscillations are always damped. However, active controls may result in poor damping of these oscillations. The DVR system utilizes limited energy storage in the form of capacitors that serve as DC energy storage; also, the series injection transformer has leakage inductances which may result in a resonance phenomenon and transient overvoltages. Detailed investigation of the harmonic resonance phenomenon in a DVR protecting a load with power factor correcting capacitors, in the perspective of control methodology of the DVR, has been carried out in Ref. [9]. In the application in Ref. [10], the shunt capacitor banks are switched on the load side of the DVR. As an alternative to the load side placement, an analysis of the supply side placement of capacitor banks is offered in this paper. Finally, complete content and organizational editing before formatting. Please take note of the following items when proofreading spelling and grammar. A. Placement of shunt capacitors on the load side of DVR Most often, the shunt capacitors are connected as power factor correction devices to a load that requires a higher power quality. The use of shunt capacitors for power factor correction introduces certain problems in the distribution system. The effects of the shunt capacitors may be felt if a power quality enhancement device such a DVR is also connected to the particular load as shown in Fig. 4. Capacitors are often connected in shunt to provide a leading current component in order to improve (increase) lagging power factor. The installed capacitors may produce oscillatory transients in the distribution system due to resonant inter- 221 www.ijaegt.com
magnification of the transient response on the load voltage when (b) capacitor banks are switched (c) DVR is switched on Fig. 4. DVR set up with capacitor bank placed on the load side. Fig. 6. Waveform of current through the DVR (load side of DVR series transformer) when capacitor banks are placed on load side Fig. 7. (a) Transient response and waveform of load voltage when capacitor banks are placed on the load side with If the foregoing remarks are accepted, it is appropriate to discuss at this point the basic advantages and disadvantages of capacitor placement near a DVR. The key concern in placing the capacitor banks on the supply side is the increase in the magnitude of the current flowing through the DVR. Depending on where reactive power revenue metering is done, the placement of capacitors on the load or supply side of the DVR may impact the customer billing. Placement of capacitors on the supply side of the DVR offers virtual isolation of control signals as mentioned above, and the power factor correction in the supply may be improved as compared to the load side placement. V. CASE STUDIES OF CAPACITOR PLACEMENTS NEAR A DVR In this section, the annotations above are further illustrated by simulation studies. The basic configuration is that of Fig. 4. The load served at the customer is 11.62 MVA, 0.86 lagging power factor. A capacitor bank of 1.8 MVAr rated reactive power is installed on the 12.47 kv side near the load side to improve the load power factor to 0.92 lagging. The result of this installation is the resonance phenomenon that causes transient overvoltage s during capacitor switching and the subsequent transient responses when the DVR is energized. A case s t u d y is performed on the simulation model to emulate the dynamic performance of the DVR and the capacitor bank. The test procedure for the case study is as follows: (1) Shunt capacitor banks are switched at t = 0.14 s; (2) At t = 0.2 s, three phase voltage sag of depth 34% is introduced on the supply side similar to the one shown in Fig. 7; and (3) At t = 0.35 s, three phase voltage sag is cleared on the supply side of the system. The transient and steady state responses of the DVR when the capacitor banks are placed on the load s i d e a n d on the supply side a r e analyzed. In the first case, the load voltage waveform and the current through the DVR are obtained when the capacitor banks are placed on the load side. Fig. 6 illustrates the waveform of the current through the DVR. The load voltage waveform is de- picted in Fig. 7a. Magnified snapshots of the transient response to the capacitor switching (t = 0.14 s) and 222 www.ijaegt.com
the firing on (t = 0.2 s) and off of the DVR (t = 0.35 s) are represented in Fig. 7b d, respectively. In the second case, the load voltage waveform and the current through the DVR are obtained when the capacitor banks are placed on the 69 kv (supply) side of the system. The load voltage waveform is depicted in Fig. 8a. Magnified snapshots of the transient response to the capacitor switching (t = 0.14 s) and the firing on (t = 0.2 s) and off of the DVR (t = 0.35 s) are represented in Fig. 8b d, respectively. Fig. 9 illustrates the waveform of the current through the DVR. Phase I load 60 Hz component (A) Table: 2 THD (%) placed on the load side RMS (A) A 649.2 3.79 656.34 B 647.9 2.46 650.92 C 648.5 5.02 660.98 placed on the supply side A 688.1 0.22 688.12 B 688.4 0.22 688.42 C 688.1 0.23 688.12 Phase V load 60 Hz component (V) THD (%) placed on the load side RMS (V) A 9301 1.56 9304.27 B 9301 1.38 9301.95 C 9301 3.03 9303.59 placed on the supply side A 9202 1.05 9202.55 B 9206 0.78 9206.30 C 9202 1.19 9202.71 Fig. 9. Waveform of current through Ann based DVR when capacitor banks are placed on supply side Table 2 represents the values of the magnitudes of the fundamental components, the total harmonic distortion and RMS values of each phase of the load voltage and current when the capacitor banks are placed on the load and the supply side. On analyzing the transient responses of the load voltage, it is observed that the overvoltages reach magnitudes of almost 18 kv when the capacitor bank is placed on the 12.47 kv side near the load as shown in Fig. 7b. The load current waveform is distorted and appears to have transient overshoots when the DVR is switched on and off (Fig. 6). As discussed previously, it is seen from Fig. 9 and Table 2 that the current through the DVR and the load is increased when the capacitor banks are sited on the supply side. 223 www.ijaegt.com
VI. CONCLUSION In this papers Ann based simulation model of DVR has been placed. The model is used to discuss the problems of v o l ta g e s we ll a nd sa g bank placement in the locality of a DVR. The compensations and drawback of capacitor location on the load or supply side h a v e been debated. It is decided that assignment of capacitors on the load s i d e r e s u l t s in lower DVR current and therefore maximal use of the DVR as a resource. The paper presents the analysis of DV R wi t h Ar ti ficia l ne ur al ne t wo r k s a nd s ho ws t he i mp ro v ed po we r q u alit y. REFERENCES [1] G.T. Heydt, Electric Power Quality, second ed., Stars in a Circle Publications, Scottsdale, 1991., p. 131. [2] G.T. Heydt, W. Tan, T. LaRose, M. Negley, Simulation and analysis of series voltage boost technology for power quality enhancement, IEEE Trans. Power Deliver. 13 (4) (1998) 1335 1341. [3] The Mathworks MATLAB and Simulink for Technical Computing, 2008. Available from: <http://www.mathworks.com/>. [4] N. Mohan, T.M. Undeland, W.P. Robbins, Power Electronics: Converters, Applications, and Design, second ed., John Wiley & Sons, New York, 1995., pp.471 475. [5] J.G. Nielsen, F. Blaabjerg, Comparison of system topologies for dynamic voltage restorers, in: Proceedings of the IEEE Industry Applications Conference 2001, vol. 1, pp. 2397 2403. [6] J.G. Nielsen, F. Blaabjerg, N. Mohan, Control strategies for dynamic voltage restorer compensating voltage sags with phase jumps, in: Proceedings on the IEEE Applied Power Electronics Conference and Exposition 2001, vol. 2, pp. 1267 1273. [7] J.G. Nielsen, M. Newman, H. Nielsen, F. Blaabjerg, Control and testing of a DVR at medium voltage, IEEE Trans. Power Electron. 19 (3) (2004) 806 813. [8] M. Vilathgamuwa, H.M. Wijekoon, S.S. Choi, A novel technique to compensate voltage sags in multiline distribution system the interline dynamic voltage restorer, IEEE Trans. Ind. Electron. 53 (5) (2006) 1603 1611. [9] M. Vilathgamuwa, H.M. Wijekoon, S.S. Choi, Investigation of the resonance phenomena in a DVR protecting a load with PF correcting capacitors, in: Proceedings of the 5th International Conference on Power Electronics and Drive Systems, vol. 1, 2003, pp. 811 815. [10] D.M. Vilathgamuwa, A.A.D.R. Perera, S.S. Choi, Voltage sag compensation with energy optimized dynamic voltage restorer, IEEE Trans. Power Deliver.18 (3) (2003) 928 936. 224 www.ijaegt.com